SEPTEMBER 2018 S H I M A D A A N D H O R I N O U C H I 2799

Reintensification and Eyewall Formation in Strong Shear: A Case Study of (2015)

UDAI SHIMADA Meteorological Research Institute, Tsukuba, Ibaraki,

TAKESHI HORINOUCHI Faculty of Environmental Earth Science, University, Sapporo, Hokkaido, Japan

(Manuscript received 27 January 2018, in final form 9 July 2018)

ABSTRACT

Strong vertical wind shear produces asymmetries in the eyewall structure of a (TC) and is generally a hostile environment for TC intensification. Typhoon Noul (2015), however, reintensified and 2 formed a closed eyewall despite 200–850-hPa vertical shear in excess of 11 m s 1. Noul’s reintensification and eyewall formation in strong shear were examined by using Doppler radar and surface observations. The evolution of the azimuthal-mean structure showed that the tangential wind at 2-km altitude increased from 30 2 to 45 m s 1 in only 5 h. During the first half of the reintensification, the azimuthal-mean inflow penetrated into the ;40-km radius, well inside the radius of maximum wind (RMW), at least below 4-km altitude, and re- flectivity inside the RMW increased. As for the asymmetric evolution, vigorous convection, dominated by an azimuthal wavenumber-1 asymmetry, occurred in the downshear-left quadrant when shear started to increase and then moved upshear. A mesovortex formed inside the convective asymmetry on the upshear side. The direction of vortex tilt between the 1- and 5-km altitudes rotated cyclonically from the downshear-left to the upshear-right quadrant as the vortex was vertically aligned. In conjunction with the alignment, the amplitude of the wavenumber-1 convective asymmetry decreased and a closed eyewall formed. These features are consistent with the theory that a vortex can be vertically aligned through upshear precession. The analysis results suggest that the vortex tilt, vigorous convection, and subsequent intensification were triggered by the increase in shear in a convectively favorable environment.

1. Introduction weakening. Explanations for the negative influence of shear on intensity include the following: (i) ventilation Tropical cyclones (TCs) form, intensify, and weaken of the TC warm core and high potential vorticity (PV) in the presence of vertical wind shear, and shear is one of by a relatively strong upper-level environmental wind the most important factors accounting for changes in (Frank and Ritchie 2001); (ii) strong downdrafts bring- the structure and intensity of a TC. The interaction ing midlevel low equivalent potential temperature (u ) processes between a TC and shear, however, are very e air into the boundary layer, leading to a decrease in TC complicated and difficult to predict. intensity from the perspective of a Carnot cycle heat Vertical shear is known to be unfavorable for in- engine (Riemer et al. 2010); and (iii) intrusion of mid- tensification. Statistically, intensifying TCs have weaker level dry air into the eyewall (Tang and Emanuel 2010). shear than nonintensifying TCs (Kaplan and DeMaria In addition, the effect of shear on intensity change is 2003). Further, shear predictors in statistical dynamical dependent not only on the shear magnitude, but also on models have a weakening effect on intensity (DeMaria vortex strength: stronger vortices are more resilient to and Kaplan 1994, 1999; Knaff et al. 2005; DeMaria 2009; shear (DeMaria 1996; Riemer et al. 2013; Rios-Berrios Kaplan et al. 2010). Paterson et al. (2005) and Wang and Torn 2017). et al. (2015) showed statistically that a 200–850-hPa 2 Jones (1995) and Reasor et al. (2004) theorized that wind shear of greater than ;10 m s 1 contributes to such TC vortices have intrinsic resiliency against shear that keeps them vertically aligned. A tilted PV column can Corresponding author: Udai Shimada, [email protected] be decomposed into azimuthal-mean and wavenumber-1

DOI: 10.1175/MWR-D-18-0035.1 Ó 2018 American Meteorological Society. For information regarding reuse of this content and general copyright information, consult the AMS Copyright Policy (www.ametsoc.org/PUBSReuseLicenses). Unauthenticated | Downloaded 10/04/21 11:42 AM UTC 2800 MONTHLY WEATHER REVIEW VOLUME 146 components (Reasor and Montgomery 2001). When the initiated by low-level deep inflow in the downshear-right scale of the vortex is much smaller than the Rossby ra- quadrant. Intense updrafts grow hydrometeors down- dius of deformation, the wavenumber-1 asymmetry be- shear, causing radar reflectivity to be highest in the haves like a vortex Rossby wave (VRW) called a tilt downshear-left quadrant. Downdrafts, which are asso- mode (Reasor et al. 2004), which is similar to an edge ciated with small-scale deep convection downshear left, wave on a Rankine vortex (Lamb 1932). The vortex dominate the upper levels of the upshear-left quadrant. precesses as the VRW propagates (Jones 1995) and is Eyewall convection on the upshear side is suppressed vertically aligned by the resonant damping of the tilt by upper-level inflow and low-level outflow. mode, provided that the sign of the radial gradient of Some studies, however, have shown that moderate PV at the critical radius, where the phase speed of the and strong shear2 can contribute to intensification VRW and tangential wind speed are the same, is nega- through the occurrence of vigorous deep convection, tive (Schecter et al. 2002; Schecter and Montgomery called convective bursts, inside the radius of maximum 1 2003). In this case, the vortex reaches an equilibrium tilt wind (RMW) on the downshear side (Molinari et al. of 908 to the left of shear (Reasor et al. 2004). In contrast, 2006; Reasor et al. 2009; Molinari and Vollaro 2010; the vortex shears apart when the gradient at the critical Nguyen and Molinari 2012). Molinari and Vollaro (2010) radius is positive (Schecter et al. 2002). Meanwhile, examined the intensification (a pressure fall of 22 hPa in when the PV gradient is 0 at the critical radius, the less than 3 h) of Tropical Storm Gabrielle (2001) in the 2 vortex undergoes repetitive cycles of tilting downshear, presence of a shear of 13 m s 1, and Nguyen and cyclonic precession upshear, and realignment (Reasor Molinari (2012) examined the intensification (an in- et al. 2004; Reasor and Montgomery 2015). Reasor et al. crease in the maximum 1-min sustained wind speed from 2 (2013) showed through a composite analysis of obser- ;33.4 to 48.9 m s 1 in 18 h) of Hurricane Irene (1999) vations that TC vortices tend to be tilted downshear and in an environment of increasing shear, from 6–7 to 2 to the left of the shear vector. 10–13 m s 1. Both Molinari and Vollaro (2010) and Previous studies have shown that maximum upward Nguyen and Molinari (2012) discussed that these storms motion occurs in the downtilt direction, rather than in were highly asymmetric, but that it was the projection of the downshear direction (Rogers et al. 2003; Zhu et al. the asymmetric heating onto the axisymmetric compo- 2004; Wu et al. 2006; Braun et al. 2006; Braun and Wu nent that was critical for intensification, based on the 2007; Ueno 2008; Riemer et al. 2010; Reasor and Eastin theoretical studies of Nolan and Grasso (2003), and 2012; Reasor et al. 2013), and mechanisms have been Nolan et al. (2007). Meanwhile, Reasor et al. (2009) proposed for this vertical motion asymmetry. When a showed that Hurricane Guillermo (1997) intensified vortex is tilted by shear, adiabatic upward (downward) rapidly with asymmetric strong convection in the motion is induced downtilt (uptilt) as a transient re- 2 downshear-left quadrant under a shear of ;7–8 m s 1 sponse to maintain thermal wind balance in the tilted and emphasized the nonnegligible contribution of eddy vortex, and this response causes a negative (positive) momentum fluxes. temperature anomaly downtilt (uptilt) (Jones 1995; Wang Even under the same moderate-shear environment, and Holland 1996; Frank and Ritchie 1999; Ueno 2007). both intensifying and nonintensifying TCs are observed. Following this response, diabatic processes allow a Rogers et al. (2016) examined structural changes of balance to be established between horizontal advec- Hurricane Edouard (2014) and showed that Edouard tion of vorticity by low-level storm-relative inflow had deep vigorous convection in the upshear-left (outflow) and vortex stretching (shrinking) on the low- quadrant during intensification but little deep convec- level inflow (outflow) side (Bender 1997; Frank and tion there during its maturity. Rios-Berrios and Torn Ritchie 1999, 2001). (2017) demonstrated, through a composite analysis of Through the mechanisms described above, vertical shear induces TC precipitation and convection structure intensifying and nonintensifying TCs under moderate with wavenumber-1 asymmetry (e.g., Black et al. 2002; shear conditions, that midlevel relative humidity and Corbosiero and Molinari 2002, 2003; Corbosiero et al. surface latent heat fluxes were larger in intensifying 2006; Chen et al. 2006). Observation-based composite TCs, in particular, on the upshear side, than in studies (Reasor et al. 2013; DeHart et al. 2014) have shown nonintensifying TCs. We lack sufficient knowledge, that for hurricane-strength TCs, eyewall convection is

2 Here, we use three categories of 200–850-hPa shear based on the 25th and 75th percentiles of shear magnitude distributions 2 1 Tilt is defined as the difference between the upper- and the around TCs (Rios-Berrios and Torn 2017): weak, ,4.5 m s 1; 2 2 lower-level vortex center locations. moderate, 4.5–11.0 m s 1; and strong, .11.0 m s 1.

Unauthenticated | Downloaded 10/04/21 11:42 AM UTC SEPTEMBER 2018 S H I M A D A A N D H O R I N O U C H I 2801

FIG. 1. (a) Ocean temperatures at 50-m depth (color scale) on 11 May 2015 and Noul’s track. Light blue dots show the center location at 6-h intervals until 0600 UTC 11 May and at 3-h intervals after that time. Black dots show the center location at 1-day intervals (0000 UTC). The large dark blue circle centered at Ishigaki Island denotes the range of the Ishigaki Doppler radar, ;200 km. The ocean temperature data were provided by the JMA. (b) Time evolution of Noul’s intensity: maximum wind (green line) and central pressure (blue line). RSMC Tokyo’s best track data do not include maximum wind speeds for a TC categorized as a tropical depression. The vertical orange bar indicates the period focused on in this study (1000–1800 UTC 11 May 2015). however, of the causes of inner-core structure and in- We describe the data and the methods used in section tensity differences in actual TCs under moderate–strong 2.Insection 3, we present an overview of Noul and as- shear conditions. One reason may be insufficient obser- sociated environmental conditions. In section 4,we vational studies: because these processes occur over the document the axisymmetric structure of Noul. In sec- ocean on a short time scale, even aircraft reconnaissance tions 5 and 6, we show the evolution of convective-scale cannot necessarily observe them at the appropriate time. structures and vortex tilt. We discuss features associ- Typhoon Noul (2015) appeared to reintensify as it ated with Noul’s reintensification in section 7. Section 8 passed near Ishigaki Island, Okinawa, Japan, from 1200 presents the conclusions. to 1800 UTC 11 May 2015 (Fig. 1a). Although this re- intensification was not captured in the best track data 2. Data, methods, and accuracy evaluation of the Regional Specialized Meteorological Center (RSMC) Tokyo (Fig. 1b), vigorous convection occurred, Data used in this study consisted of surface observa- and then the eyewall structure changed from asymmet- tions provided by the JMA, C-band operational Doppler ric to symmetric in the presence of shear greater than radar data observed at Ishigaki Island (Fig. 1a), radar 2 11 m s 1. We present evidence for this reintensification composite rainfall data provided by the JMA, infrared in section 3a. (IR) brightness temperatures (at 10.4 mm) from the Why did Noul reintensify and acquire a closed eyewall Himawari-8 geostationary satellite, and Japanese structure despite strong shear? We use high-temporal- 55-year Reanalysis (JRA-55) atmospheric data (Kobayashi resolution surface observations and Japan Meteorolog- et al. 2015). The surface observations included maxi- ical Agency (JMA) operational Doppler radar data to mum gust wind, temperature, dewpoint temperature (or address this question. We aim to document the evolu- relative humidity), precipitation, station pressure, and tion of the inner-core structure of Noul, and to examine sea level pressure (SLP). Here, the maximum gust wind the processes involved in Noul’s reintensification and is the maximum 3-s mean wind speed and direction the formation of a closed eyewall in a strong-shear observed during each 1-min period. Equivalent poten- environment. tial temperature ue was calculated by using the formula

Unauthenticated | Downloaded 10/04/21 11:42 AM UTC 2802 MONTHLY WEATHER REVIEW VOLUME 146

FIG. 2. (a) Noul’s track. The black line shows the track obtained by the GBVTD-simplex method at 5-min intervals; the black dots are at 1-h intervals. The light blue line shows RSMC Tokyo’s best track, and the light blue dot is the storm position at 1500 UTC 11 May. (b) Doppler velocity plan position indicator (PPI) pattern at a 1.18 elevation angle at 1330 UTC 11 May. The black dot indicates the center detected in this study, the red dot indicates the center location in RSMC Tokyo’s best track data, and the black circle denotes the 1-km RMW. of Bolton (1980). The Doppler observation parameters the retrieval area, however, when the TC was within of the JMA radars have been reported by Shimada et al. 60 km of the radar site. This condition introduced some (2016, their Table 1). uncertainty into the center-finding method and the re- By applying the ground-based velocity track display sulting uncertainty of center locations is a limitation of (GBVTD) technique (Lee et al. 1999) to Doppler ve- this study. The storm center, defined as the center de- locity, VD, data, we retrieved the two-dimensional hor- tected at 1-km altitude, was used in the GBVTD analysis izontal wind field at 1-km altitude intervals from 1 to to retrieve the axisymmetric component of the wind 10 km of altitude for Noul. No boundary layer flow was field at all altitudes. During the period of the wind retrieved because of the lack of VD data. The dataset dataset gap, the center locations were obtained by includes tangential winds with wavenumbers 0, 1, 2, temporal interpolation. To examine the vertical tilt of and 3, and radial winds with wavenumbers 0 and 1 in Noul, we also detected the storm center at 5-km altitude, storm-centered cylindrical coordinates. Details of the using the same method that we used at 1-km altitude. GBVTD-derived wind dataset are given by Shimada Centers above 5-km altitude could not be reliably de- et al. (2018). The analysis period was from 1200 to tected because the Doppler velocity data were too 1700 UTC 11 May 2015. The GBVTD technique can sparse to obtain reasonable results. retrieve the wind field only within the range of radii Noul’s track, as detected by the GBVTD-simplex between the storm center and the radar location. Be- method, was displaced a few tens of kilometers west- cause no VD observations could be made in the re- ward from RSMC Tokyo’s best track (Fig. 2a). At gion, no winds were retrieved while the eye was over the 1330 UTC 11 May, as the TC approached the radar site radar site from 1405 to 1450 UTC 11 May. on Ishigaki Island, the detected center location was in In the GBVTD technique, the storm center is defined the middle of the newly formed eyewall and in an ap- as the location that maximizes the axisymmetric tan- propriate location for the VD pattern. In contrast, the gential wind y on the scale of the RMW. To detect the best track center location, obtained by interpolation of storm center, we used the GBVTD-simplex center- the best track data, was located near the eyewall where 21 finding method (Lee and Marks 2000; Bell and Lee VD was ;20 m s (Fig. 2b). We consider that this result 2012), in which the time continuity of the RMW, maxi- justifies our use of the detected track in this study. mum y, and the center location is used to find the center As the metric to evaluate the quality of the wind data among candidate locations (for more details, see Bell retrieved by the GBVTD technique, we used the overall and Lee 2012). As the likely scale of the RMW, the average of the root-mean-square difference (RMSD) range of radii from 30 to 60 km was set in the simplex between the VD resampled from the GBVTD-retrieved method. The RMW scale was not sufficiently covered in winds and the observed VD. The RMSDs, which generally

Unauthenticated | Downloaded 10/04/21 11:42 AM UTC SEPTEMBER 2018 S H I M A D A A N D H O R I N O U C H I 2803

FIG. 3. (a) Time evolutions of the overall average RMSD between VD resampled from the GBVTD-retrieved winds at altitudes from 1 to 5 km and observed VD at altitudes from 1 to 5 km. (b) Doppler velocity PPI pattern at a 0.28 elevation angle at 1621 UTC 11 May. The black circle denotes a radius of 30 km as representative of the 1-km RMW. The small black arrows show gust wind vectors observed at around 1621 UTC, the bold red and blue arrows highlight the VD directions, and the dashed ellipse marks a convergence area in the VD field.

2 ranged from 3 to 6 m s 1 (Fig. 3a), were larger values increasing, and ending at 1800 UTC 11 May, a few hours than those obtained by other case studies (Zhao et al. before the central pressure started to increase again 2012; Shimada et al. 2018). The relatively poor quality of (Fig. 1b). At 1000 UTC 11 May, convective bursts oc- the GBVTD-retrieved winds is likely to stem from curred to the north (downshear-left quadrant) of Noul sparse radar coverage in the first half of the analysis (Figs. 4a,b). After Noul passed near Ishigaki Island, the period (e.g., Fig. 2b), noise contamination due to VD coldest IR brightness temperature was observed on the aliasing (e.g., Fig. 3b), and the poor accuracy of the TC upshear-left side, but there was no well-defined eye center-finding method (e.g., Lee and Marks 2000; (Fig. 4c). Harasti et al. 2004; Bell and Lee 2012; Murillo et al. Following the convective bursts, GBVTD-retrieved 2011). The distortion of the dipole VD pattern associated y at altitudes of 1 and 2 km increased rapidly from ;30 2 with the inbound and outbound Doppler velocities to ;45 m s 1 during only 5 h from 1200 to 1700 UTC (Fig. 3b), caused by the complicated flow, can also reduce 11 May (Fig. 5). This increase in y is evidence of Noul’s the accuracy of the GBVTD-retrieved winds. Thus, the reintensification. After 1545 UTC, y was almost the GBVTD-retrieved wind speeds have some uncertainty. same at 1- and 2-km altitudes because the number of For this reason, we did not use GBVTD-retrieved asym- plan position indicator (PPI) scans from the Ishigaki metric wind data in this study; instead, we used maximum radar was limited when Noul was far from the radar gust wind data observed at weather stations. site, with the result that the constant-altitude plan po- sition indicator (CAPPI) data, which were used in the 3. Overview of Typhoon Noul GBVTD analysis, were almost the same at 1- and 2-km altitudes. a. Storm history and intensity b. Environmental conditions According to the best track data of RSMC Tokyo (Fig. 1), Typhoon Noul attained a lifetime minimum Around the onset of reintensification (1200 UTC), central pressure of 920 hPa at 0000 UTC 10 May 2015 Noul was passing over the ocean south of Ishigaki over the . Then, after passing just north- Island, where temperatures at 50-m depth were greater east of Island, it weakened rapidly. Noul was than 278C(Fig. 1a); then, after Noul passed near Ishigaki transformed into an at 0300 UTC Island, ocean temperatures at 50-m depth along the 12 May. For more details of Noul, see the 2015 annual track decreased to below 258C. 2 report of the JMA (2016). Noul was under a westerly shear of ;11 m s 1 around In this study, we focus on the period beginning at the onset of reintensification, but by 1800 UTC the shear 2 1000 UTC 11 May, a few hours before the central had increased to more than 16 m s 1 (Fig. 6). The for- pressure in the best track data temporarily stopped ward speed of Noul was very fast; after 1300 UTC, it was

Unauthenticated | Downloaded 10/04/21 11:42 AM UTC 2804 MONTHLY WEATHER REVIEW VOLUME 146

FIG. 4. Infrared brightness temperatures (at 10.4 mm) from the Himawari-8 geostationary satellite at (a) 1030, (b) 1200, and (c) 1640 UTC 11 May. The 3 mark at the center of each panel indicates the approximate center of Noul, which was subjectively determined from radar imagery. The white arrow shows the 850–200-hPa shear vector.

2 greater than 10 m s 1, and after 1700 UTC it exceeded increase in reflectivity inside the RMW and the low- 2 20 m s 1. The gradual increases in shear and forward level inflow in the vicinity of the RMW are generally speed were caused by the influence of an upper-level jet favorable conditions for an increase in maximum y associated with a deep trough over mainland (not (Shapiro and Willoughby 1982; Pendergrass and shown; discussed further in section 7b). Willoughby 2009; Nolan et al. 2007; Smith et al. 2009; Rogers et al. 2013; Smith and Montgomery 2016). The accuracy of the GBVTD-retrieved winds appears 4. Axisymmetric structure poorer during period II than period I because the distri- The axisymmetric structure of Noul changed dramati- butions of both y and u during period II were noisy cally during the 5 h (1200–1700 UTC 11 May) of the radar (Figs. 7a,b); y fluctuated erratically, and inflow and outflow analysis. Hereafter, 1200–1400 UTC 11 May, before the were mixed from 1530 to 1630 UTC (Fig. 7b). The 1-km gap in the wind dataset, is defined as period I of the RMW was sometimes within 20–25 km of the center, where analysis, and 1500–1700 UTC, after the gap, is defined as the retrieved y was anomalously large (Fig. 7a). period II. During period I, the axisymmetric structure of Noul at 1-km altitude shows that isopleths of the azimuthal- mean absolute angular momentum, M [ (1/2)fr2 1 ry, where f is the Coriolis parameter and r is radius, moved inward with time as the radial gradient of M increased (Fig. 7a); this result indicates that both y and absolute z [ › › vorticity, a (1/r)( M/ r), increased. In addition, the RMW at 1-km altitude contracted from ;65 km at 1200 UTC to ;40 km by 1510 UTC (Fig. 7a). The vortex itself, however, was still weak and Noul did not yet 2 have a well-organized eyewall. The 30 m s 1 contour of y and the 10-dBZ contour of radar reflectivity were at low altitudes and had wide radial widths (Fig. 8a). Below 4-km altitude, the inflow penetrated into a radius of 40 km, well inside the RMW (Figs. 7b and 8b), and outside the RMW the inflow at 1-km altitude exceeded 2 24ms 1 (Fig. 7b). From 1315 to 1430 UTC, azimuthal- mean reflectivity increased, and the radius of maximum FIG. 5. Time evolutions of maximum azimuthal-mean tangential reflectivity (RMR) was inside the RMW (Fig. 7c). The wind speed at altitudes of 1 and 2 km.

Unauthenticated | Downloaded 10/04/21 11:42 AM UTC SEPTEMBER 2018 S H I M A D A A N D H O R I N O U C H I 2805

5. Convective-scale structure a. Convective bursts Noul’s reintensification began with convective bursts. The rainfall imagery shows that, around the onset of reintensification (1200 UTC), intense convection was located downshear left, ;75 km from the storm center (Fig. 9a). This convection approached the storm center, playing a role in the contraction of the RMW (Figs. 9b,c), and moved cyclonically around the storm center from the downshear to the upshear side as it became organized into an eyewall (Figs. 9d,e). The convection weakened, but did not dissipate when it moved to the upshear side of the eyewall, where downward motion generally domi- nates the lower troposphere of TCs in shear (Reasor et al. FIG. 6. Vertical wind shear (black line) and its heading direction (red line) between 200 and 850 hPa averaged within 500 km from 2013; DeHart et al. 2014). When the convection was ac- the storm center. The shear was calculated from JRA-55 data tive primarily inside the 5-km RMW, from 1315 to (Kobayashi et al. 2015). The shaded bar indicates the radar obser- 1600 UTC (Figs. 9b,c,d), y increased rapidly at the 1- and vation period. 2-km altitudes (Fig. 5). Finally, by 1800 UTC, a closed eyewall had formed (Fig. 9f). In spite of the uncertainty in y at 1-km altitude, y What caused the convective bursts? Sonde observa- definitely increased until at least 1600 UTC. During tions at Ishigaki Island, which was located in the 2 period II, the 30 m s 1 contour of y and the 10-dBZ downshear-left quadrants of Noul until ;1200 UTC contour of reflectivity reached, on average, an altitude 11 May, showed that the ue averaged over the lowest of 6 and 10 km, respectively (Figs. 8c,d). Azimuthal- 500 m greatly exceeded the saturation equivalent po- mean reflectivity increased outside the RMW after tential temperature, ues, at altitudes from 2 to 10 km at 1500 UTC, and the 20-dBZ contour outside the RMR 0600 UTC before the start of the convective bursts moved outward with time (Fig. 7c). (Fig. 10). This sonde profile indicates the potential for

5 2 21 FIG. 7. (a) Radius–time Hovmöller diagram of y at 1-km altitude (color scale), M (10 m s , purple contours), and RMW (thick black line). The blank areas are where the GBVTD technique could not retrieve y (i.e., during the TC’s passage near Ishigaki Island). (b) Radius–time Hovmöller diagram of azimuthal mean radial wind u at 1-km altitude (color scale) and RMW at 1-km altitude (thick black line). The blank areas are where the GBVTD technique could not retrieve u. (c) Radius–time Hovmöller diagram of azimuthal mean radar reflectivity at 1-km altitude (color), radius of maximum reflectivity (RMR; red line), and RMW (thick black line). The black horizontal lines indicate the period I and II boundaries.

Unauthenticated | Downloaded 10/04/21 11:42 AM UTC 2806 MONTHLY WEATHER REVIEW VOLUME 146

21 FIG. 8. Radius–height plots of (a) time-averaged y (m s , black contours), azimuthal-mean radar reflectivity 2 2 (color scale), and M (105 m2 s 1, red contours); and (b) time-averaged u (m s 1, black contours positive, white 2 contours negative), azimuthal-mean radar reflectivity (color scale), and M (105 m2 s 1, red contours) during period I. The dashed line is the RMW. Contours were drawn only in areas where there were observations averaged over at least 1 h in total. (c),(d) As in (a),(b), but for period II. upright buoyant convection. Because storm-relative ra- storm center at four specific times (e.g., plot periods dial winds at the surface in the downshear-left quadrant A–D) (Fig. 11, Table 1). The SR winds were derived were inflow (see section 5b), the convective bursts likely from surface gust wind observations. occurred in a convectively favorable environment. During plot periods A–D, inflow was occurring on the downshear side and outflow on the upshear side of b. Mesovortex the storm. This radial flow pattern is generally observed In association with the convective bursts, a meso- in the lower troposphere of TCs in shear (e.g., Braun vortex formed in the eye. To visualize the mesovortex, et al. 2006; Reasor et al. 2013). During plot period A 2 we plotted storm-relative (SR) surface winds at 4-min (Fig. 11a), SR wind speed was ;30 m s 1 around 20 km intervals at weather station locations relative to the northwest of the storm center inside the eyewall. During

Unauthenticated | Downloaded 10/04/21 11:42 AM UTC SEPTEMBER 2018 S H I M A D A A N D H O R I N O U C H I 2807

FIG. 9. (a)–(f) Radar composite rainfall rate (data provided by the JMA). In each panel, the bold black arrow shows the 850–200-hPa shear vector; the black dot indicates the radar site, and the black (red) circle denotes the 1-km (5 km) RMW. The 5-km RMW was obtained by using the 5-km centers in the GBVTD analysis. The center locations at 1- and 5-km altitudes in (c) were obtained by temporal interpolation; as a result, the relative center locations are not necessarily accurate. The 1-km RMW in (c) and the 5-km RMWs in (b), (c), and (d) were also obtained by temporal interpolation. Note that the rainfall distribution, in particular that on the north side of Noul, has been artificially affected by radar attenuation. plot period B (Fig. 11b), a concave portion, along the time series obtained at Nakasuji station on Tarama inner edge of the eyewall, which was ;20 km northeast Island (Fig. 12). These time series show the inner-core of the storm center during period A, was northwest of structure of Noul at the surface from downshear to the center and inside an active convective asymmetry upshear, because during period II Noul was moving 2 with reflectivity greater than 40 dBZ that extended from northeastward with a speed greater than 18 m s 1 and southwest to northeast. In the concave portion, SR wind it passed very near Tarama Island. From 1527 to 2 speed was weak, below 10 m s 1, and a potential cyclonic 1531 UTC, when the mesovortex made its nearest ap- circulation was forming at around 10 km northwest of the proach to Tarama Island (Fig. 10c), the sign of the SR storm center. From period A to period B, SR wind speed tangential winds ySR changed (Fig. 12a); in addition, ue increased on the left side of the track, except in the concave reached ;368 K, and SLP fell to ;981 hPa (Fig. 12c). portion. During period C, a small-scale cyclonic circulation The locations of the SLP minimum and the high-ue peak was found southwest of the storm center. Hereafter, we were displaced from the storm center by 15–20 km to- refer to this local circulation as a mesovortex. During pe- ward the active convective asymmetry (Fig. 12b). The riod D, the mesovortex moved counterclockwise along the radial profiles of ue and SLP within 35 km from the storm inner edge of the eyewall (Fig. 11d). center were distorted from symmetry, suggesting that The mesovortex was characterized by a SLP minimum the scale of the mesovortex was ;15 km, much smaller and a high-ue peak, as seen in the surface observation than the RMW (;27 km) and the radius of maximum

Unauthenticated | Downloaded 10/04/21 11:42 AM UTC 2808 MONTHLY WEATHER REVIEW VOLUME 146

side could not have been realized without the advection

of high-ue air from the eye region by the mesovortex. Guimond et al. (2016) reported, on the basis of air- craft observations of Hurricane Karl (2010), a similar configuration of a mesovortex in the eye and convective bursts in the eyewall. They showed that convective bursts occurred in an area of low-level convergence of outflow with warm buoyant air associated with meso- vortices in the eye and vortex-scale flow. Eastin et al. (2005) reported, on the basis of aircraft observations, the origin of buoyant updrafts in the eyewall appeared to be

high-ue air advected outward by mesovortices from the eye, where ue was much (;10 K) greater than that in the eyewall. Reasor et al. (2009) speculated that the mixing of warm, moist air from the eye offset the negative in- fluence of relatively strong shear and led to the rapid intensification (RI) of Hurricane Guillermo (1997). In Noul, relatively strong SR tangential winds of 2 greater than 30 m s 1 were present at the inner edge of the convection (Figs. 12a,b). Using the SR tangential winds observed at Tarama Island and the following FIG. 10. Vertical profiles of the difference between the equiva- formula for relative vorticity: u lent potential temperature e averaged over the lowest 500 m ›y y u SR SR and the saturated equivalent potential temperature es at each z 5 1 , altitude from sonde observations at Ishigaki Island (see Fig. 2a). ›r r The warm colors denote vertical levels where the parcel’s ue below 500 m exceeds the surrounding environment’s ues. Con- we calculated that relative vorticity at the inner edge 22 21 4 vective available potential energy (CAPE), which is calculated of the convection reached 0.8 3 10 s . Near the 2 by lifting a parcel averaged over the lowest 500 m, is shown on mesovortex, a maximum gust wind speed of 58.6 m s 1 the top of the panel. (earth-relative wind) was observed at Shimoji (Figs. 2a and 11d). This very strong wind was also confirmed 21 rainfall (;40 km). From 1533 to 1541 UTC, just prior to by the VD of greater than 60 m s at 1–2-km altitude the passage of the active convective asymmetry over (Fig. 3b). A similar very strong wind along the inside

Nakasuji (Fig. 12b), a steep horizontal gradient of ue was edge of the eyewall was also observed in Hurricane 2 observed, reaching 15 K (10 km) 1, in the area of out- Hugo (1989) (Marks et al. 2008).

flow (Fig. 12a), indicative of positive ue advection from the eye into the convection. 6. Vortex tilt At 1621 UTC, the PPI pattern at an elevation angle of 21 0.28 showed that the 0 m s isopleth of VD was nearly a. Wavenumber-1 reflectivity asymmetry perpendicular to the radar range direction in the vicinity Wavenumber-0–75 reflectivities averaged over radii of of the convective asymmetry on the west side of Noul 25–50 km (around the contracting RMW and outside the (Fig. 3b, dashed oval).3 This orientation implies the ex- RMW) showed that a wavenumber-1 reflectivity asym- istence of horizontal convergence along the convective metry with amplitude greater than 10 dBZ was present area. The V height in the convergence area in Fig. 3b is D until 1625 UTC (Fig. 13). The radial range in Fig. 13 was 1–2 km. This configuration, that is, positive u advection e determined from the dominant region of wavenumber-1 inside the eyewall and convergence along the convective asymmetry from 1330 to 1600 UTC (Figs. 14a–d); during area, was favorable for the maintenance of the convec- this period, a positive wavenumber-1 anomaly propagated tion upshear. Because SR outflow was observed radially outside the convection upshear (Fig. 11), we speculate that the maintenance of the convection on the upshear 4 Note that the calculated vorticity results from the superposition of Noul primary vortex and the mesovortex. 5 Note that since the storm center was defined as the center at 3 Note that VD is the Earth-relative wind velocity, not the SR 1-km altitude, the reflectivity asymmetry is shown with respect to wind velocity. the vortex at 1-km altitude.

Unauthenticated | Downloaded 10/04/21 11:42 AM UTC SEPTEMBER 2018 S H I M A D A A N D H O R I N O U C H I 2809

21 21 FIG. 11. Distribution of storm-relative (SR) surface winds (small arrows: red $ 36 m s and black , 36 m s ) and composite radar reflectivity (color scale) at (a) 1330, (b) 1430, (c) 1520, and (d) 1610 UTC 11 May. SR winds observed at each weather station during each period shown in Table 1 are plotted at their location relative to the TC center position at each specific time. The black line is the track determined by the GBVTD-simplex method until 1700 UTC 11 May. The red ellipse indicates the area of active asymmetric convection on which this study focused. The bold black arrow shows the 850–200-hPa shear vector. The black circle denotes the 1-km RMW. The dashed arrow circle indicates the concave portion of the inner edge of the eyewall. The brown arrow circle indicates the location of a possible mesovortex. The 1-km RMW indicated in (b) was obtained by a temporal interpolation. cyclonically from the downshear-left side to the up- and 50 km rapidly decreased (Fig. 13), and there was shear side at an almost single phase speed between no significant wavenumber-1 asymmetry within a radii of 25 and 50 km. The propagation speed is dis- 70-km radius, except around the 20-km radius, where the cussed further in section 6c. After 1600 UTC, the deviation from a closed eyewall caused an asymmetry wavenumber-1 asymmetry between the radii of 25 (Fig. 14e). Instead, wavenumber-0 reflectivity rapidly

TABLE 1. Plot periods for the analysis of storm-relative (SR) surface winds at four specific times (Fig. 11). Note that not all SR winds observed at each weather station are plotted during the plot periods. If a weather station was located in the eye during any two plot periods, then SR winds in the eye were plotted during the period closer to the observed time.

Specific times Plot periods Location of the convective asymmetry A 1330 UTC 11 May 1230–1430 UTC North-northeast of Noul B 1430 UTC 11 May 1330–1530 UTC Northwestern side of Noul C 1520 UTC 11 May 1450–1600 UTC Western side of Noul D 1610 UTC 11 May 1530–1640 UTC Southwestern side of Noul

Unauthenticated | Downloaded 10/04/21 11:42 AM UTC 2810 MONTHLY WEATHER REVIEW VOLUME 146

FIG. 12. Time evolutions of surface observations at Nakasuji (location shown in Fig. 2a). 21 (a) Storm-relative surface tangential wind velocity ySR (m s , black line) and radial wind 21 velocity uSR (m s , red line). (b) Distance between the storm-scale center and the weather 2 station (km, black line) and surface rainfall (mm h 1, blue line). (c) Equivalent potential temperature ue (K, orange line) and sea level pressure (SLP) (hPa, black dashed line). The dashed vertical lines refer to the distance from the storm center shown in (b). increased with time, becoming dominant by 1700 UTC GBVTD analysis was poor, particularly during period (Fig. 13). II. According to Rogers et al. (2003) and Reasor et al. (2013), both the upward motion maximum and the tilt b. Tilt evolution direction tend to be found upstream of the rainfall We examined the evolution of vortex tilt in relation to maximum. In Noul, the azimuthal location of the vortex the wavenumber-1 reflectivity asymmetry. The tilt di- tilt coincided with that of the mesovortex during period rection of the inner-core vortex (on the scale of the II (Fig. 15). The center locations at 5-km altitude de- RMW) between the center at 1-km altitude and that at tected by the center-finding algorithm were systemati- 5-km altitude was almost the same as that of the positive cally displaced toward the location of the mesovortex. wavenumber-1 reflectivity from 1230 to 1330 UTC The tilt precession might have been partly due to the (Fig. 15). Note that before 1330 UTC the positive cyclonic propagation of the mesovortex because the 2 wavenumber-1 asymmetry was mainly found in the very strong wind speeds greater than 50 m s 1, which are downshear-left quadrant outside the 50-km radius suggestive of the mesovortex, were observed even at (Fig. 14a). After 1330 UTC the tilt was 458–908 down- 5-km altitude (not shown). stream of the wavenumber-1 reflectivity, reaching the The tilt magnitude was relatively large (;5–15 km) upshear-right quadrant, though with erratic fluctuations when the wavenumber-1 reflectivity was strong (1200– (Fig. 15). Note, however, that there exists uncertainty in 1600 UTC) (Figs. 16a,b). After 1600 UTC, however, the analysis of the vortex tilt because the quality of the when the vortex tilted upshear, the tilt magnitude rapidly

Unauthenticated | Downloaded 10/04/21 11:42 AM UTC SEPTEMBER 2018 S H I M A D A A N D H O R I N O U C H I 2811

Previous studies have shown that maximum upward motion occurs downtilt (e.g., Braun et al. 2006; Ueno 2008) and that the amplitude of the wavenumber-1 up- ward motion asymmetry in the eyewall is strongly cor- related with the amplitude of the wavenumber-1 rainfall asymmetry (Ueno 2007). Therefore, it is plausible that the formation of the closed eyewall of Noul was asso- ciated with the alignment of the vortex after 1600 UTC. c. Comparison with theory We further examined the vortex tilt in the context of a tilt mode of VRWs in shear (Reasor et al. 2004; Reasor and Montgomery 2015). The wavenumber-1 asymmetry

FIG. 13. Time evolution of wavenumber-0–7 radar reflectivity was primarily observed between radii of 25 and 50 km averaged over radii of 25–50 km (simple averages were computed (Fig. 13), where the radial gradient of relative vorticity without considering the areal mean). The reflectivity is averaged at 1-km altitude was negative (Fig. 17). This negative between 1- and 2-km altitudes. radial gradient would allow VRWs to propagate upwind relative to the tangential wind, under the assumption decreased to below 4 km; this reduction in the magni- that the radial gradient of vorticity has the same sign as tude indicates that the vortex was nearly aligned after the radial gradient of PV. In fact, the wavenumber-1 its upshear precession from the downshear-left quad- convective asymmetry at a radius of 37.5 km propagated 2 rant in strong shear (Fig. 16a). This alignment occurred cyclonically at an earth-relative speed of 8.9 m s 1, at the same time as the drop in the amplitude of which was ;24% of the corresponding y at 1-km altitude 2 the wavenumber-1 reflectivity asymmetry (Fig. 16b). (37.2 m s 1, averaged over 25–50-km radius) during the

FIG. 14. Wavenumeber-1 radar reflectivity distribution within a 100-km radius from the storm center at (a) 1300, (b) 1430, (c) 1515, (d)1600,and(e)1700UTC.Thereflectivityis averaged between 1- and 2-km altitudes. The black large arrow shows the 850–200-hPa shear vector in each panel. The thin black circles are at radii of 5, 20, 40, 60, and 80 km from the center and the thick green circles are at radii of 25 and 50 km from the center.

Unauthenticated | Downloaded 10/04/21 11:42 AM UTC 2812 MONTHLY WEATHER REVIEW VOLUME 146

radius, which was located outside the analysis domain (Fig. 17). If damping of the tilt mode is absent, the vortex can be aligned through its precession upshear; this process matches Noul’s actual behavior. Reasor and Montgomery (2015) derived a heuristic model of a tilt mode without damping in cylindrical coordinates (r, u, z, t), where u is the azimuthal angle (counterclockwise), z is the pseudoheight vertical coordinate (Hoskins and Bretherton 1972), and t is time, as follows:     v t v t p  0 u ’22U dq p u 2 p z q (r, , z, t) v sin cos cos , p dr 2 2 H FIG. 15. Time–azimuthal distribution of wavenumber-1 reflectivity averaged between radii of 25 and 50 km, and between 1- and 2-km (1) altitudes. The thick black line indicates the direction of the vortex tilt determined from the difference in the center location between 1- and where q and q0 are the axisymmetric and asymmetric 5-km altitudes. The slanting lines divide the shear quadrants. ‘‘DL,’’ components of PV, respectively; U is half the magnitude ‘‘UL,’’ ‘‘UR,’’ and ‘‘DR’’ denote downshear-left, upshear-left, of the vertical shear; v is the vortex precession fre- upshear-right, and downshear-right quadrants, respectively. The p green stars indicate the direction of the mesovortex from the quency; and H is the depth of the vortex. The tilt is storm center. maximized at u 5 p/2 (908 to the left of shear), and the tilt magnitude between the surface and half of H can 0 0 be estimated as jr j [ jq /(dq/dr)j 5 j2U/vpj. The vortex period from 1330 to 1630 UTC. This fact may allow the precession frequency vp is expressed as double the fre- observed wavenumber-1 asymmetry to be interpreted quency of the observed wavenumber-1 VRW, 2 3 (V/R), as a tilt mode of VRWs. where V is the observed wavenumber-1 VRW propaga- 2 To what degree is the observed, possible tilt mode tion speed (;9ms 1) and R is the mean radius of the 2 consistent with theory? The behavior of the tilt mode wave (;37.5 km). Since vertical shear was ;13 m s 1 at 2 depends on the radial gradient of PV at the critical ra- 1500 UTC, U is 6.5 m s 1. Substituting these values into 0 dius, as described in the introduction. Unfortunately, we j2U/vpj yields jr j 5;27 km. If it is assumed that H is could not confirm the sign of the gradient at the critical ;12 km, which is equivalent to the height of ;200-hPa

FIG. 16. Time evolution of (a) tilt direction and (b) magnitude determined from differences in the center location between 1- and 5-km altitudes. In (a), the origin of the plot is the storm center at 1-km altitude and the displacement of the 5-km center relative to the 1-km center is drawn.

Unauthenticated | Downloaded 10/04/21 11:42 AM UTC SEPTEMBER 2018 S H I M A D A A N D H O R I N O U C H I 2813

precession upshear, and the formation of a closed eye- wall in an environment of increasing shear caused by an approaching upper-level trough. Here, these features are compared with features of other intensifying TCs in strong shear. a. Intensification in strong shear Some previous studies have suggested that strong shear can favorably contribute to TC intensification through processes such as downshear reformation or asymmetric intensification. Nguyen and Molinari (2015) showed through a numerical simulation of Tropical Storm Gabrielle (2001) that during RI, an upright, intense mesovortex formed downshear by vorticity stretching in a sufficiently moist environment, and that the mesovortex FIG. 17. Radial profiles of time-averaged (from 1330 to 1630 UTC) azimuthal-mean relative vorticity (red line, left axis) and y at 1-km evolved into the TC vortex as its flow axisymmetrized the altitude (black line, right axis). After the time averaging, the rel- original vortex. Meanwhile, Musgrave et al. (2008) per- ative vorticity values were radially filtered using a 5-km running formed numerical experiments of Gabrielle and showed mean. The vertical dotted lines indicate the radial range from 25 that when vertical shear was reduced, the simulated in- to 50 km. tensity was weaker than the actual intensity. None of these cases of asymmetric intensification, however, was accom- level (according to sonde observations at Ishigaki Island), panied by a closed eyewall. and that the tilt magnitude increases linearly with height, In Noul, similar to other intensifying TCs, the timing then the tilt magnitude between 1- and 5-km altitudes of the increase in y was consistent with the timing of the (Dz 5 4km) is estimated to be (2jr0jDz)/H 5;18 km. increase in azimuthal-mean reflectivity inside the RMW This estimate is much larger than the actual tilt mag- (Fig. 7c). This result suggests that the increase in the nitude (;7.5 km) at around 1500 UTC (Fig. 16b). A axisymmetric component of diabatic heating through possible explanation for this difference is that the local asymmetric convection was important for Noul’s re- shear in the inner core of the vortex was much less than intensification, as described by Molinari and Vollaro the estimated value of U. In addition, processes such as (2010) and Nguyen and Molinari (2012). The convective axisymmetrization (e.g., Enagonio and Montgomery bursts initially occurred in a favorable thermodynamic 2001), diabatic PV production (DeMaria 1996; Davis environment in the downshear-left quadrant (sections et al. 2008), and stretching, and upward advection, of 5a). The tilt direction of the inner-core vortex was also in vorticity (Nguyen and Molinari 2015; Miyamoto and the downshear-left quadrant at that time (section 6b). If Nolan 2018), may have decreased the tilt magnitude. adiabatic upward motion was forced on the downtilt side After 1700 UTC, the eyewall structure remained nearly in response to increasing shear (Jones 1995; Ueno 2007), symmetric for a few hours (Fig. 9f). This fact suggests it is plausible that the increase in shear around the onset that damping of the tilt mode occurred. If so, the magni- of reintensification caused Noul to tilt, and triggered the tude of the downshear-left equilibrium tilt can be esti- convective bursts. When the convection propagated 0 0 mated as jr j [ jq /(dq/dr)j 5 jU/vpj [Eq. (14) of Reasor upshear, the advection of high-ue air from the eye by the and Montgomery (2015)]. By 1700 UTC, vertical shear was mesovortex likely contributed to the maintenance of the 21 ;15 m s .Forvp, y is assumed to have increased by convection (section 5b). 2 ;5ms 1 (Fig. 5), and this increase leads to an increase in With regard to the vertical alignment during in- 2 V (14 m s 1). Here R is assumed to have contracted to tensification, it might be useful to consider Hurricane ;30 km as the RMW contracted. Thus, the tilt magnitude Earl (2010) because, similar to Noul, Earl became ver- between 1- and 5-km altitudes is estimated to be ;5km. tically aligned after being significantly tilted to 908 left of This large decrease in the theoretical tilt magnitude is shear (Stevenson et al. 2014; Rogers et al. 2015; Susca- consistent with the decrease in the observed tilt magnitude. Lopata et al. 2015). Earl was characterized by a displace- ment of ;50 km between circulation centers at 2- and 8-km altitudes prior to RI in moderate shear. Over the 7. Discussion next 12 h, RI and the vertical alignment of the vortex Noul’s reintensification was characterized by convec- occurred in conjunction with convective bursts inside tive bursts, vertical alignment of the vortex following its the RMW. Chen and Gopalakrishnan (2015) showed

Unauthenticated | Downloaded 10/04/21 11:42 AM UTC 2814 MONTHLY WEATHER REVIEW VOLUME 146 through numerical simulation that Earl was aligned after Ishigaki Island and operational Doppler radar data pro- the onset of RI, a result that is consistent with the vided an opportunity to document Noul’s reintensification. alignment of Noul during its reintensification. An im- We therefore examined why Noul reintensified and why it portant difference, however, is that Earl continued to had a closed eyewall in strong shear. rapidly intensify into a major hurricane, whereas Noul The evolution of the azimuthal-mean structure showed began to weaken again shortly afterward because of low that, in association with the occurrence of convective ocean temperatures (Fig. 1a) and strong shear more bursts, inflow penetrated into a radius of ;40 km, well 2 than 16 m s 1 (Fig. 6). inside the RMW, at least below 4-km altitude, and azimuthal-mean reflectivity inside the RMW increased. b. Effect of an upper-level trough Further, the maximum y at 2-km altitude rapidly in- 2 It has been shown that TCs can intensify through in- creased from 30 to 45 m s 1 during only 5 h and the 1-km teraction with an upper-level trough and associated RMW contracted from ;65 to ;40 km. These features jet (Molinari and Vollaro 1990; DeMaria et al. 1993; are consistent with those of intensifying TCs reported by Hanley et al. 2001), though a more recent study by previous studies. Peirano et al. (2016) cast doubt on the positive influence The convective bursts were dominated by a positive of troughs on TCs. An upper-level jet preferably trans- wavenumber-1 asymmetry in the downshear-left quad- fers negative angular momentum outward to spin up the rant. The asymmetry between the radii of 25 and 50 km outflow layer of the TC. The eddy momentum flux was organized into the eyewall and rotated cyclonically convergence (EFC) is defined as follows: to the upshear side. In conjunction with the convective asymmetry, the direction of the vortex tilt between 1 › 0 0 EFC [2 r2u y , (2) 1- and 5-km altitudes rotated cyclonically from the r2 ›r SR SR downshear-left to the upshear-right quadrant while the where uSR is the storm-relative radial wind, the overbar tilt magnitude decreased to less than 4 km. The ampli- indicates the azimuthal mean, and the primes indicate the tude of the wavenumber-1 convective asymmetry also deviation from the azimuthal mean. To indicate a trough decreased and a closed eyewall formed. These features interaction objectively, Hanley et al. (2001) used as an indicate that after the vortex precessed cyclonically into indicator an EFC averaged over a 300–600-km radial range the upshear-right quadrant in strong shear, the vortex 2 2 at the 200-hPa level that exceeds 10 m s 1 day 1 con- was vertically aligned. This behavior is qualitatively tinuously for more than 12 h. consistent with the theory that TC vortices can resist An upper-level jet axis associated with a deep trough tilting by vertical shear. was located ;1500 km north of Noul when the convec- Surface observations showed that a mesovortex oc- tive bursts started at around 1000 UTC 11 May (not curred inside the convective asymmetry on the upshear shown). We calculated EFC at each radius by using side. The mesovortex was characterized by warm moist JRA-55 reanalysis data and found that the radius where air and a pressure minimum, and the horizontal gradient 21 21 u it exceeded 10 m s day was ;850 km from Noul’s of e between the mesovortex and the convective 21 center as of 1200 UTC (not shown); thus, the trough was asymmetry reached 15 K (10 km) . The configuration too far away to spin up Noul. Therefore, from the of the mesovortex was thus favorable for the advection u perspective of EFC, we could not conclude that Noul of high- e air from the eye to the active convective re- reintensified through interaction with the upper-level gion on the upshear side of the storm. trough. The results of our analysis suggest that the vortex tilt, convective bursts, and subsequent intensification were triggered by the increase in vertical shear in a convectively 8. Conclusions favorable environment. To our knowledge, this study is Typhoon Noul (2015) underwent reintensification and the first one based on high-temporal-resolution obser- developed a closed eyewall during its passage near Ish- vations of a TC that was vertically aligned as it precessed igaki Island from 1000 to 1800 UTC 11 May 2015. upshear during intensification. Although many obser- During this period, westerly to southwesterly vertical vational studies have examined the relationships among 2 wind shear increased from 11 to 16 m s 1 and the speed shear, an asymmetric structure, and intensity change, of Noul’s northeastward movement increased from 10 few studies have examined the dramatic changes of a TC 2 to 20 m s 1. The evolution of Noul during this period vortex occurring on a short time scale, ;6 h, in response was atypical, because strong vertical shear is generally to an increase in shear. Future studies should examine unfavorable for intensification and tends to induce an further the effects of shear on TCs on the basis of high- asymmetric eyewall structure. Surface observations around temporal-resolution observations and simulations.

Unauthenticated | Downloaded 10/04/21 11:42 AM UTC SEPTEMBER 2018 S H I M A D A A N D H O R I N O U C H I 2815

Acknowledgments. U. Shimada is deeply grateful to DeMaria, M., 1996: The effect of vertical shear on tropical cyclone Dr. R. Oyama, Mr. H. Hara, Dr. M. Sawada, and intensity change. J. Atmos. Sci., 53, 2076–2088, https://doi.org/ , . Prof. H. Niino for fruitful discussions. The authors 10.1175/1520-0469(1996)053 2076:TEOVSO 2.0.CO;2. ——, 2009: A simplified dynamical system for tropical cyclone in- thank three anonymous reviewers for valuable com- tensity prediction. Mon. Wea. Rev., 137, 68–82, https://doi.org/ ments that have greatly improved the manuscript. The 10.1175/2008MWR2513.1. opinions in this paper are those of the authors and ——, and J. Kaplan, 1994: A Statistical Hurricane Intensity Pre- should not be regarded as official RSMC Tokyo views. diction Scheme (SHIPS) for the Atlantic basin. Wea. Fore- casting, 9, 209–220, https://doi.org/10.1175/1520-0434(1994) 009,0209:ASHIPS.2.0.CO;2. REFERENCES ——, and ——, 1999: An updated Statistical Hurricane Intensity Prediction Scheme (SHIPS) for the Atlantic and eastern Bell, M. M., and W. Lee, 2012: Objective tropical cyclone center North Pacific basins. Wea. Forecasting, 14, 326–337, https:// tracking using single-Doppler radar. J. Appl. Meteor. Clima- doi.org/10.1175/1520-0434(1999)014,0326:AUSHIP.2.0.CO;2. tol., 51, 878–896, https://doi.org/10.1175/JAMC-D-11-0167.1. ——, J.-J. Baik, and J. Kaplan, 1993: Upper-level eddy angular Bender, M. A., 1997: The effect of relative flow on the asymmetric momentum flux and tropical cyclone intensity change. structure in the interior of hurricanes. J. Atmos. Sci., 54, J. Atmos. Sci., 50, 1133–1147, https://doi.org/10.1175/1520- , 703–724, https://doi.org/10.1175/1520-0469(1997)054 0703: 0469(1993)050,1133:ULEAMF.2.0.CO;2. . TEORFO 2.0.CO;2. Eastin, M. D., W. M. Gray, and P. G. Black, 2005: Buoyancy of Black, M. L., J. F. Gamache, F. D. Marks, C. E. Samsury, and H. E. convective vertical motions in the inner core of intense hur- Willoughby, 2002: Eastern Pacific Hurricanes Jimena of 1991 ricanes. Part II: Case studies. Mon. Wea. Rev., 133, 209–227, and Olivia of 1994: The effect of vertical shear on structure https://doi.org/10.1175/MWR-2849.1. and intensity. Mon. Wea. Rev., 130, 2291–2312, https://doi.org/ Enagonio, J., and M. T. Montgomery, 2001: Tropical cyclogene- 10.1175/1520-0493(2002)130,2291:EPHJOA.2.0.CO;2. sis via convectively forced vortex Rossby waves in a shal- Bolton, D., 1980: The computation of equivalent potential tem- low water primitive equation model. J. Atmos. Sci., 58, perature. Mon. Wea. Rev., 108, 1046–1053, https://doi.org/ 685–706, https://doi.org/10.1175/1520-0469(2001)058,0685: 10.1175/1520-0493(1980)108,1046:TCOEPT.2.0.CO;2. TCVCFV.2.0.CO;2. Braun, S. A., and L. Wu, 2007: A numerical study of Hurricane Frank, W. M., and E. A. Ritchie, 1999: Effects of environmental Erin (2001). Part II: Shear and the organization of eyewall flow upon tropical cyclone structure. Mon. Wea. Rev., 127, vertical motion. Mon. Wea. Rev., 135, 1179–1194, https:// 2044–2061, https://doi.org/10.1175/1520-0493(1999)127,2044: doi.org/10.1175/MWR3336.1. EOEFUT.2.0.CO;2. ——, M. T. Montgomery, and Z. Pu, 2006: High-resolution simu- ——, and ——, 2001: Effects of vertical wind shear on the intensity lation of Hurricane Bonnie (1998). Part I: The organization of and structure of numerically simulated hurricanes. Mon. Wea. eyewall vertical motion. J. Atmos. Sci., 63, 19–42, https:// Rev., 129, 2249–2269, https://doi.org/10.1175/1520-0493(2001) doi.org/10.1175/JAS3598.1. 129,2249:EOVWSO.2.0.CO;2. Chen, H., and S. G. Gopalakrishnan, 2015: A study on the asym- Guimond, S. R., G. M. Heymsfield, P. D. Reasor, and A. C. metric rapid intensification of Hurricane Earl (2010) using the Didlake, 2016: The rapid intensification of Hurricane Karl HWRF system. J. Atmos. Sci., 72, 531–550, https://doi.org/ (2010): New remote sensing observations of convective bursts 10.1175/JAS-D-14-0097.1. from the Global Hawk platform. J. Atmos. Sci., 73, 3617–3639, Chen, S., J. A. Knaff, and F. D. Marks Jr., 2006: Effects of vertical https://doi.org/10.1175/JAS-D-16-0026.1. wind shear and storm motion on tropical cyclone rainfall Hanley, D., J. Molinari, and D. Keyser, 2001: A composite of the in- asymmetries deduced from TRMM. Mon. Wea. Rev., 134, teractions between tropical cyclones and upper-tropospheric 3190–3208, https://doi.org/10.1175/MWR3245.1. troughs. Mon. Wea. Rev., 129, 2570–2584, https://doi.org/ Corbosiero, K. L., and J. Molinari, 2002: The effects of vertical 10.1175/1520-0493(2001)129,2570:ACSOTI.2.0.CO;2. wind shear on the distribution of convection in tropical cy- Harasti, P. R., C. J. McAdie, P. P. Dodge, W.-C. Lee, J. Tuttle, S. T. clones. Mon. Wea. Rev., 130, 2110–2123, https://doi.org/ Murillo, and F. D. Marks, 2004: Real-time implementation of 10.1175/1520-0493(2002)130,2110:TEOVWS.2.0.CO;2. single-Doppler radar analysis methods for tropical cyclones: ——, and ——, 2003: The relationship between storm motion, Algorithm improvements and use with WSR-88D display vertical wind shear, and convective asymmetries in tropical data. Wea. Forecasting, 19, 219–239, https://doi.org/10.1175/ cyclones. J. Atmos. Sci., 60, 366–376, https://doi.org/10.1175/ 1520-0434(2004)019,0219:RIOSRA.2.0.CO;2. 1520-0469(2003)060,0366:TRBSMV.2.0.CO;2. Hoskins, B. J., and F. P. Bretherton, 1972: Atmospheric fronto- ——, ——, A. R. Aiyyer, and M. L. Black, 2006: The structure genesis models: Mathematical formulation and solution. and evolution of Hurricane Elena (1985). Part II: Convec- J. Atmos. Sci., 29, 11–37, https://doi.org/10.1175/1520-0469(1972) tive asymmetries and evidence for vortex Rossby waves. 029,0011:AFMMFA.2.0.CO;2. Mon. Wea. Rev., 134, 3073–3091, https://doi.org/10.1175/ JMA, 2016: Annual report on the activities of the RSMC Tokyo- MWR3250.1. Typhoon Center. Japan Meteorological Agency, 103 pp., Davis, C. A., S. C. Jones, and M. Riemer, 2008: Hurricane vortex http://www.jma.go.jp/jma/jma-eng/jma-center/rsmc-hp-pub-eg/ dynamics during Atlantic extratropical transition. J. Atmos. AnnualReport/2015/Text/Text2015.pdf. Sci., 65, 714–736, https://doi.org/10.1175/2007JAS2488.1. Jones, S. C., 1995: The evolution of vortices in vertical shear. I: DeHart, J. C., R. A. Houze, and R. F. Rogers, 2014: Quadrant Initially barotropic vortices. Quart. J. Roy. Meteor. Soc., 121, distribution of tropical cyclone inner-core kinematics in re- 821–851, https://doi.org/10.1002/qj.49712152406. lation to environmental shear. J. Atmos. Sci., 71, 2713–2732, Kaplan, J., and M. DeMaria, 2003: Large-scale characteristics of https://doi.org/10.1175/JAS-D-13-0298.1. rapidly intensifying tropical cyclones in the North Atlantic

Unauthenticated | Downloaded 10/04/21 11:42 AM UTC 2816 MONTHLY WEATHER REVIEW VOLUME 146

basin. Wea. Forecasting, 18, 1093–1108, https://doi.org/ Part II: Symmetric response and nonlinear simulations. J. Atmos. 10.1175/1520-0434(2003)018,1093:LCORIT.2.0.CO;2. Sci., 60, 2717–2745, https://doi.org/10.1175/1520-0469(2003) ——, ——, and J. A. Knaff, 2010: A revised tropical cyclone rapid 060,2717:NTPTBH.2.0.CO;2. intensification index for the Atlantic and eastern North Pacific ——, Y. Moon, and D. P. Stern, 2007: Tropical cyclone in- basins. Wea. Forecasting, 25, 220–241, https://doi.org/10.1175/ tensification from asymmetric convection: Energetics and ef- 2009WAF2222280.1. ficiency. J. Atmos. Sci., 64, 3377–3405, https://doi.org/10.1175/ Knaff, J. A., C. R. Sampson, and M. DeMaria, 2005: An operational JAS3988.1. statistical typhoon intensity prediction scheme for the western Paterson, L. A., B. N. Hanstrum, N. E. Davidson, and H. C. Weber, North Pacific. Wea. Forecasting, 20, 688–699, https://doi.org/ 2005: Influence of environmental vertical wind shear on 10.1175/WAF863.1. the intensity of hurricane-strength tropical cyclones in the Kobayashi, S., and Coauthors, 2015: The JRA-55 Reanalysis: Australian region. Mon. Wea. Rev., 133, 3644–3660, https:// General specifications and basic characteristics. J. Meteor. doi.org/10.1175/MWR3041.1. Soc. Japan, 93, 5–48, https://doi.org/10.2151/jmsj.2015-001. Peirano, C. M., K. L. Corbosiero, and B. H. Tang, 2016: Revisiting Lamb, H., 1932: Hydrodynamics. 6th ed. Dover, 732 pp. trough interactions and tropical cyclone intensity change. Lee, W.-C., and F. D. Marks Jr., 2000: Tropical cyclone kinematic Geophys. Res. Lett., 43, 5509–5515, https://doi.org/10.1002/ structure retrieved from single-Doppler radar observations. 2016GL069040. Part II: The GBVTD-simplex center finding algorithm. Mon. Pendergrass, A. G., and H. E. Willoughby, 2009: Diabatically in- Wea. Rev., 128, 1925–1936, https://doi.org/10.1175/1520- duced secondary flows in tropical cyclones. Part I: Quasi- 0493(2000)128,1925:TCKSRF.2.0.CO;2. steady forcing. Mon. Wea. Rev., 137, 805–821, https://doi.org/ ——, B. J.-D. Jou, P.-L. Chang, and S.-M. Deng, 1999: Tropical 10.1175/2008MWR2657.1. cyclone kinematic structure retrieved from single-Doppler Reasor, P. D., and M. T. Montgomery, 2001: Three-dimensional radar observations. Part I: Interpretation of Doppler velocity alignment and corotation of weak, TC-like vortices via linear patterns and the GBVTD technique. Mon. Wea. Rev., 127, vortex Rossby waves. J. Atmos. Sci., 58, 2306–2330, https:// 2419–2439, https://doi.org/10.1175/1520-0493(1999)127,2419: doi.org/10.1175/1520-0469(2001)058,2306:TDAACO.2.0.CO;2. TCKSRF.2.0.CO;2. ——, and M. D. Eastin, 2012: Rapidly intensifying Hurricane Marks, F. D., P. G. Black, M. T. Montgomery, and R. W. Burpee, Guillermo (1997). Part II: Resilience in shear. Mon. Wea. Rev., 2008: Structure of the eye and eyewall of Hurricane Hugo 140, 425–444, https://doi.org/10.1175/MWR-D-11-00080.1. (1989). Mon. Wea. Rev., 136, 1237–1259, https://doi.org/ ——, and M. T. Montgomery, 2015: Evaluation of a heuristic model 10.1175/2007MWR2073.1. for tropical cyclone resilience. J. Atmos. Sci., 72, 1765–1782, Miyamoto, Y., and D. Nolan, 2018: Structural changes preceding https://doi.org/10.1175/JAS-D-14-0318.1. rapid intensification in tropical cyclones as shown in a large ——, ——, and L. D. Grasso, 2004: A new look at the problem of ensemble of idealized simulations. J. Atmos. Sci., 75, 555–569, tropical cyclones in vertical shear flow: Vortex resiliency. https://doi.org/10.1175/JAS-D-17-0177.1. J. Atmos. Sci., 61,3–22,https://doi.org/10.1175/1520-0469(2004) Molinari, J., and D. Vollaro, 1990: External influences on hurricane 061,0003:ANLATP.2.0.CO;2. intensity. Part II: Vertical structure and response of the hur- ——, M. D. Eastin, and J. F. Gamache, 2009: Rapidly intensifying ricane vortex. J. Atmos. Sci., 47, 1902–1918, https://doi.org/ Hurricane Guillermo (1997). Part I: Low-wavenumber struc- 10.1175/1520-0469(1990)047,1902:EIOHIP.2.0.CO;2. ture and evolution. Mon. Wea. Rev., 137, 603–631, https:// ——, and ——, 2010: Rapid intensification of a sheared tropical doi.org/10.1175/2008MWR2487.1. storm. Mon. Wea. Rev., 138, 3869–3885, https://doi.org/ ——, R. Rogers, and S. Lorsolo, 2013: Environmental flow impacts 10.1175/2010MWR3378.1. on tropical cyclone structure diagnosed from airborne Dopp- ——, P. Dodge, D. Vollaro, K. L. Corbosiero, and F. Marks, 2006: ler radar composites. Mon. Wea. Rev., 141, 2949–2969, https:// Mesoscale aspects of the downshear reformation of a tropical doi.org/10.1175/MWR-D-12-00334.1. cyclone. J. Atmos. Sci., 63, 341–354, https://doi.org/10.1175/ Riemer, M., M. T. Montgomery, and M. E. Nicholls, 2010: A new JAS3591.1. paradigm for intensity modification of tropical cyclones: Murillo, S. T., W.-C. Lee, M. M. Bell, F. D. Marks Jr., P. P. Dodge, Thermodynamic impact of vertical wind shear on the inflow and G. M. Barnes, 2011: Intercomparison of Ground-Based layer. Atmos. Chem. Phys., 10, 3163–3188, https://doi.org/ Velocity Track Display (GBVTD)-retrieved circulation cen- 10.5194/acp-10-3163-2010. ters and structures of Hurricane Danny (1997) from two ——, ——, and ——, 2013: Further examination of the thermo- coastal WSR-88Ds. Mon. Wea. Rev., 139, 153–174, https:// dynamic modification of the inflow layer of tropical cyclones doi.org/10.1175/2010MWR3036.1. by vertical wind shear. Atmos. Chem. Phys., 13, 327–346, Musgrave, K. D., C. A. Davis, and M. T. Montgomery, 2008: Nu- https://doi.org/10.5194/acp-13-327-2013. merical simulations of the formation of Hurricane Gabrielle Rios-Berrios, R., and R. D. Torn, 2017: Climatological analysis of (2001). Mon. Wea. Rev., 136, 3151–3167, https://doi.org/ tropical cyclone intensity changes under moderate vertical 10.1175/2007MWR2110.1. wind shear. Mon. Wea. Rev., 145, 1717–1738, https://doi.org/ Nguyen, L. T., and J. Molinari, 2012: Rapid intensification of a 10.1175/MWR-D-16-0350.1. sheared, fast-moving hurricane over the Gulf Stream. Mon. Rogers, R. F., S. S. Chen, J. Tenerelli, and H. E. Willoughby, 2003: Wea. Rev., 140, 3361–3378, https://doi.org/10.1175/MWR-D- A numerical study of the impact of vertical shear on the dis- 11-00293.1. tribution of rainfall in Hurricane Bonnie (1998). Mon. Wea. ——, and ——, 2015: Simulation of the downshear reformation of a Rev., 131, 1577–1599, https://doi.org/10.1175//2546.1. tropical cyclone. J. Atmos. Sci., 72, 4529–4551, https://doi.org/ ——, P. Reasor, and S. Lorsolo, 2013: Airborne Doppler obser- 10.1175/JAS-D-15-0036.1. vations of the inner-core structural differences between in- Nolan, D. S., and L. D. Grasso, 2003: Nonhydrostatic, three- tensifying and steady-state tropical cyclones. Mon. Wea. Rev., dimensional perturbations to balanced, hurricane-like vortices. 141, 2970–2991, https://doi.org/10.1175/MWR-D-12-00357.1.

Unauthenticated | Downloaded 10/04/21 11:42 AM UTC SEPTEMBER 2018 S H I M A D A A N D H O R I N O U C H I 2817

——, ——, and J. Zhang, 2015: Multiscale structure and evolution Earl (2010). Mon. Wea. Rev., 142, 4364–4380, https://doi.org/ of Hurricane Earl (2010) during rapid intensification. Mon. 10.1175/MWR-D-14-00078.1. Wea. Rev., 143, 536–562, https://doi.org/10.1175/MWR-D-14- Susca-Lopata, G., J. Zawislak, E. J. Zipser, and R. F. Rogers, 2015: 00175.1. The role of observed environmental conditions and pre- ——, J. Zhang, J. Zawislak, H. Jiang, G. Alvey, E. Zipser, and cipitation evolution in the rapid intensification of Hurricane S. Stevenson, 2016: Observations of the structure and evolu- Earl (2010). Mon. Wea. Rev., 143, 2207–2223, https://doi.org/ tion of Hurricane Edouard (2014) during intensity change. 10.1175/MWR-D-14-00283.1. Part II: Kinematic structure and the distribution of deep Tang, B., and K. A. Emanuel, 2010: Midlevel ventilation’s con- convection. Mon. Wea. Rev., 144, 3355–3376, https://doi.org/ straint on tropical cyclone intensity. J. Atmos. Sci., 67, 1817– 10.1175/MWR-D-16-0017.1. 1830, https://doi.org/10.1175/2010JAS3318.1. Schecter, D. A., and M. T. Montgomery, 2003: On the symmetri- Ueno, M., 2007: Observational analysis and numerical evaluation zation rate of an intense geophysical vortex. Dyn. Atmos. of the effects of vertical wind shear on the rainfall asymmetry Oceans, 37, 55–88, https://doi.org/10.1016/S0377-0265(03) in the typhoon inner-core region. J. Meteor. Soc. Japan, 85, 00015-0. 115–136, https://doi.org/10.2151/jmsj.85.115. ——, ——, and P. D. Reasor, 2002: A theory for the vertical ——, 2008: Effects of ambient vertical wind shear on the inner-core alignment of a quasigeostrophic vortex. J. Atmos. Sci., 59, asymmetries and vertical tilt of a simulated tropical cyclone. 150–168, https://doi.org/10.1175/1520-0469(2002)059,0150: J. Meteor. Soc. Japan, 86, 531–555, https://doi.org/10.2151/ ATFTVA.2.0.CO;2. jmsj.86.531. Shapiro, L., and H. E. Willoughby, 1982: The response of balanced Wang, Y., and G. J. Holland, 1996: The beta drift of baroclinic vorti- hurricanes to local sources of heat and momentum. J. Atmos. ces. Part I: Adiabatic vortices. J. Atmos. Sci., 53, 411–427, https:// Sci., 39, 378–394, https://doi.org/10.1175/1520-0469(1982) doi.org/10.1175/1520-0469(1996)053,0411:TBDOBV.2.0.CO;2. 039,0378:TROBHT.2.0.CO;2. ——, Y. Rao, Z.-M. Tan, and D. Schnemann, 2015: A statistical Shimada, U., M. Sawada, and H. Yamada, 2016: Evaluation of the analysis of the effects of vertical wind shear on tropical cyclone accuracy and utility of tropical cyclone intensity estimation intensity change over the western North Pacific. Mon. Wea. using single ground-based Doppler radar observations. Mon. Rev., 143, 3434–3453, https://doi.org/10.1175/MWR-D-15- Wea. Rev., 144, 1823–1840, https://doi.org/10.1175/MWR-D- 0049.1. 15-0254.1. Wu, L., S. A. Braun, J. Halverson, and G. Heymsfield, 2006: A ——, ——, and ——, 2018: Doppler radar analysis of the rapid numerical study of Hurricane Erin (2001). Part I: Model ver- intensification of (2015) after eyewall re- ification and storm evolution. J. Atmos. Sci., 63, 65–86, https:// placement. J. Atmos. Sci., 75, 143–162, https://doi.org/10.1175/ doi.org/10.1175/JAS3597.1. JAS-D-17-0042.1. Zhao, K., M. Xue, and W.-C. Lee, 2012: Assimilation of GBVTD- Smith, R. K., and M. T. Montgomery, 2016: The efficiency of diabatic retrieved winds from single-Doppler radar for short-term heating and tropical cyclone intensification. Quart. J. Roy. forecasting of super typhoon Saomai (0608) at landfall. Meteor. Soc., 142, 2081–2086, https://doi.org/10.1002/qj.2804. Quart. J. Roy. Meteor. Soc., 138, 1055–1071, https://doi.org/ ——, ——, and V. S. Nguyen, 2009: Tropical cyclone spin-up re- 10.1002/qj.975. visited. Quart. J. Roy. Meteor. Soc., 135, 1321–1335, https://doi. Zhu, T., D.-L. Zhang, and F. Weng, 2004: Numerical simulation of org/10.1002/qj.428. Hurricane Bonnie (1998). Part I: Eyewall evolution and in- Stevenson, S. N., K. L. Corbosiero, and J. Molinari, 2014: The tensity changes. Mon. Wea. Rev., 132, 225–241, https://doi.org/ convective evolution and rapid intensification of Hurricane 10.1175/1520-0493(2004)132,0225:NSOHBP.2.0.CO;2.

Unauthenticated | Downloaded 10/04/21 11:42 AM UTC